Need some sort of driver IC

[triden]

I am working on a project that has to drive 7-segment displays. However, they aren't your average displays and are about 5" long and 3.5" wide. They draw about 250mA of current when fully on. I am looking for some sort of switch/amplifier that can be driven by my BCD-7segment encoder to power a display. Right now I have one of these:

ULN2003A - SEVEN DARLINGTON ARRAYS
http://www.xdesignlabs.com/chips/uln2003a.pdf

But they only work by switching the ground. These are common cathode displays, so I need to switch the VCC side of the displays. I would have used a mosfet or BJT for each segment on the display, but I don't have the room. Can anyone recommend something? It would be awesome if it was in a DIP package. Thanks

Christan

 

[berkeman]

I did a quick google search on ULN2003 +highside, and got some hits that look to be useful for you. I added the keyword "highside" to get hits that would show you (hopefully) equivalent PNP highside versions of the NPN lowside ULN2003.

Sorry that I don't have time to go through the hits to find you a part at the moment -- I have to bail for a few hours. Good luck!

 

[triden]

Berkeman, I don't know if it's just me or what, but it looks almost impossible to find a highside darlington array driver..with atleast 7 outputs. I scoured google to no avail..maybe I should try something else? Is there another packed device that will work similarily? Maybe something like a mosfet or BJT switch package?

Thanks though!

 

[berkeman]

Hmm, yeah, when I narrow the search with +PNP, I only get a few hits (including this PF thread already!). You might be right that there aren't any highside equaivalents of the ULN2003.

But also, are you sure you want to put all 7 drivers in the same package for each digit? That's a lot of power. A Darlington's saturation voltage is close to a volt, so that would be almost 2W per package. Do you have any other package options?

 

[triden]

Other package options will work, it's just that I need to power 4 displays and having a discrete transistor for each digit will take a fair amount of space. If you know of something else I could use..that would be great! A digit draws about 30mA @ 20 volts ~ 4.2 watts per display. What would you do in this case?

 

[berkeman]

Well, I'd be inclined not to use a Darlington pair for the driver. 30mA is low enough that you can do it with a simple saturated discrete transistor to stay more efficient (0.2V to 0.3V Vsat, versus that plus an additional 0.65V for the Darlington pair).

I'd start with discrete 2N2907 transistors as the highside PNP switches, and see how the layout and size looked. Each will only be dissipating 10mW or so, so they are not very stressed. The next step in integration (though not a really big savings in PCB area) would be to look at the PNP quads in DIP packages (like the SMT SO-16). I know companies like Motorola put quad 2N3906 transistors in SO-16 packages, but I don't know if there's a good 2N2907-type quad package available. The 3906 is a little light for your 30mA, but still might be useable. Check out the 3906 specs, and see if you can find a quad 2907, which would be better.

Is there a reason that you were looking at Darlingtons for only 30mA/segment? I misunderstood initially, and thought it was 250mA/segment.

 

[NoTime]

I agree with berkeman that it sounds more like a layout problem. The amount difference of board real estate shouldn't be that large.

However, you might want to consider scanning the display.
That could decrease the component count (and space requirements) considerably.

 

[triden]

Well, I'd be inclined not to use a Darlington pair for the driver. 30mA is low enough that you can do it with a simple saturated discrete transistor to stay more efficient (0.2V to 0.3V Vsat, versus that plus an additional 0.65V for the Darlington pair).

I'd start with discrete 2N2907 transistors as the highside PNP switches, and see how the layout and size looked. Each will only be dissipating 10mW or so, so they are not very stressed. The next step in integration (though not a really big savings in PCB area) would be to look at the PNP quads in DIP packages (like the SMT SO-16). I know companies like Motorola put quad 2N3906 transistors in SO-16 packages, but I don't know if there's a good 2N2907-type quad package available. The 3906 is a little light for your 30mA, but still might be useable. Check out the 3906 specs, and see if you can find a quad 2907, which would be better.

Is there a reason that you were looking at Darlingtons for only 30mA/segment? I misunderstood initially, and thought it was 250mA/segment.

Sorry for the misunderstanding (sometimes I don't write what I think). Each segment draws 30mA which is about 210mA when the display is lit (All 7 segments on). Originally I had the ULN2003 because it could power 500mA total which would be enough for each display, but my displays are common cathode, and the ULN2003 is setup for a common anode display.

Here's a pic that might help. The 4511D (BCD to 7-segment decoder) emits a "high" to turn a segment on. Then the ULN2003A takes this signal and drives that corrosponding segment. Each display is connected to a "channel" via an ethernet wire. Since my displays are CC, I need common ground to the display instead of a common +9V. Right now, the schematic shows the ULN2003 to sink the current instead of sourcing it, unlike what I need. If I could get an IC to source the current, that would be awesome. I'm having trouble finding such an IC..without doing a discrete transistor for each segment (thats 28 transistors).

 

[NoTime]

One of these then?
http://www.allegromicro.com/sf/2981/

Alternates
http://www.allegromicro.com/ic/leddriver.asp [Broken]

 

[triden]

One of these then?
http://www.allegromicro.com/sf/2981/

Alternates
http://www.allegromicro.com/ic/leddriver.asp [Broken]

Thanks! Just what I was looking for.

 

[berkeman]

Thanks NoTime. Occam's razor strikes again!

 

[NoTime]

Thanks NoTime. Occam's razor strikes again!

:rofl: When doing design work your first solution often gets trashed by ugly reality.
Cost, size and other constraints (like availability as in this case) are big killers.
You really need to be willing and ready to revisit the drawing board.